3. 索引爆炸

  1. #  读取employee数据集，设定行索引是RACE
  2.  In[22]: employee = pd.read_csv('data/employee.csv', index_col='RACE')
  3.          employee.head()
  4. Out[22]:

3. 索引爆炸 - 图1

  1. #  选取BASE_SALARY做成两个Series，判断二者是否相同
  2.  In[23]: salary1 = employee['BASE_SALARY']
  3.          salary2 = employee['BASE_SALARY']
  4.          salary1 is salary2
  5. Out[23]: True
  1. #  结果是True，表明二者指向的同一个对象。这意味着，如果修改一个，另一个也会去改变。为了收到一个全新的数据，使用copy方法：
  2.  In[24]: salary1 = employee['BASE_SALARY'].copy()
  3.          salary2 = employee['BASE_SALARY'].copy()
  4.          salary1 is salary2
  5. Out[24]: False
  1. #  对其中一个做索引排序，比较二者是否不同
  2.  In[25]: salary1 = salary1.sort_index()
  3.          salary1.head()
  4. Out[25]: RACE
  5.          American Indian or Alaskan Native    78355.0
  6.          American Indian or Alaskan Native    26125.0
  7.          American Indian or Alaskan Native    98536.0
  8.          American Indian or Alaskan Native        NaN
  9.          American Indian or Alaskan Native    55461.0
  10.          Name: BASE_SALARY, dtype: float64
  12.  In[26]: salary2.head()
  13. Out[26]: RACE
  14.          Hispanic/Latino    121862.0
  15.          Hispanic/Latino     26125.0
  16.          White               45279.0
  17.          White               63166.0
  18.          White               56347.0
  19.          Name: BASE_SALARY, dtype: float64
  1. #  将两个Series相加
  2.  In[27]: salary_add = salary1 + salary2
  3.  In[28]: salary_add.head()
  4. Out[28]: RACE
  5.          American Indian or Alaskan Native    138702.0
  6.          American Indian or Alaskan Native    156710.0
  7.          American Indian or Alaskan Native    176891.0
  8.          American Indian or Alaskan Native    159594.0
  9.          American Indian or Alaskan Native    127734.0
  10.          Name: BASE_SALARY, dtype: float64
  1. #  再将salary1与其自身相加；查看几个所得结果的长度，可以看到长度从2000到达了117万
  2.  In[29]: salary_add1 = salary1 + salary1
  3.          len(salary1), len(salary2), len(salary_add), len(salary_add1)
  4. Out[29]: (2000, 2000, 1175424, 2000)

更多

  1. #  验证salary_add值的个数。因为笛卡尔积是作用在相同索引元素上的，可以对其平方值求和
  2.  In[30]: index_vc = salary1.index.value_counts(dropna=False)
  3.          index_vc
  4. Out[30]: Black or African American            700
  5.          White                                665
  6.          Hispanic/Latino                      480
  7.          Asian/Pacific Islander               107
  8.          NaN                                   35
  9.          American Indian or Alaskan Native     11
  10.          Others                                 2
  11.          Name: RACE, dtype: int64
  13.  In[31]: index_vc.pow(2).sum()
  14. Out[31]: 1175424